Importing without unique field

[Former Member]

Hi All,

We are into a project which deals with Harmonization of data from various systems.We extract the data from all ECC system and load them into MDM,do the matching merging and assign a unique number.

Now our problem is while loading there is no unique field value given in the source file .So what we planned is to use the combination of all fields to act as unique and load the data.The source file does not contain and mandatory and unique value fields.

After we do match merge in data manager we assign the unique number.So i wanted to know is this the correct way to use all the fields in MDM to import the data.Or is there any other solution for the same.I know that MDM performance will go down if combination of all fields is taken to import the records

Thanks and Regards

Nitin

[Former Member]

Hi All,

Thanks for the reply.Thread closed

Nitin

[Former Member]

Please paste your final solution for other's benefit.

Thanks,

Shambhu.

[Former Member]

Hi ,

Thansk for the reply.

Our model is not one customer to many adress concept but instead its other way around like one address to many customer number which are stored as different customer number in different ERP system.

Same customer can be stored with different number in different ERP system.When we import this in MDM we will have one address in main table and tuple will have 4 customer numbers with respective source system.

After matching merging it is not sent back to any of the ERP ,it is sent to BW via PI.PI comes into picture only here.

Regards

Nitin

[Former Member]

Since this a different data model and you are running import based on a combination of 30 fields, it will be better to simulate in Quality environment and do a performance testing.

Refer MDIS.ini parameters to improve the Import Server performance - http://help.sap.com/saphelp_mdm550/helpdata/en/7e/02a3705d1f465f9608d16b96d1e8a2/frameset.htm

Similarly set MDS.ini parameters - http://help.sap.com/saphelp_mdm550/helpdata/en/42/fc006b8adf1a61e10000000a422035/content.htm

Thanks,

Shambhu.

[Former Member]

Hi Nitin,

I would suggest rather than 30 fields you can still find some unique field or combination of 3-4 fields to import data for Address using rather than all 30 fields. If could possible try to pick 4-5 fields combination which can be uniqe in your scenario for example;

areaCode , localNumber, full name, Extension and Mobile phone number etc. So you need to evaluate some data for the same.

Otherwise as already suggested by others you can chek the performanace of Import Manager using these 30 fields combination and can further do silicing and chunks of data as per requirement.

Regards,

Mandeep Saini

[Former Member]

Hi All,

Thanks for the reply.

Here is the scenarion below

Client wants to combine all commercial operations worldwide and move to an integrated commercial team based on country, regional territories and global crops. Clients As-Is organization has independent operating divisions within each country or region. Each division have their own SAP ERP systems and hence this does not provide an integrated view of its customer data.

As a first step, it wants to enable existing reporting system to do reporting on integrated customer view by harmonizing customer records across multiple SAP platforms

So the cleansed data is not sent to any ERP system but instead it is sent to BW for reporting.So IDOC type wont be a factor here.Please correct me if I am wrong here.

There is a main table and a tuple.Main table contains address data and Tuple contains only the customer number and source system( like form which system it is ).So no question of key field inaddress table as none of the field has mandated value.

Secondly at first instance we load around 650000 records of cleansed records.We wont perform any matching merging on this.

Here the importing is not a problem because unique number will be provided.

Then when the delta records of around 5000 to 6000 records comes which does not have unique value field we load this using the combination of all 30 fields.

Here was the question like first we have aroung 650000 records in data manger and when we try to load the delta records with with combination of 30 fields still the performance will be stable and good or will it affect?

Now we cannot have any keyfields defined in respective ERP systems.

Please provide the guidance am i going in correct direction or do we have any other solution

Thanks and Regards

Nitin

[Former Member]

According to me the data model should be like One customer to Many Address mapping.

You should have the following in SAP MDM -

1) Customer Main Table with Remote System and Remote Key, Country, Region, Division etc..

Since you say that you have SAP ERP, then you should have unique customer ids in each ERP.

Read Remote Key concept as posted in my previous message. This key will store the source system name and the unique id in source system for the customer.

Read this guide - https://websmp210.sap-ag.de/~sapidb/012003146900000303082009E/MDM_Customer_Content_Information.pdf

2) Customer Main table should have a field linked to the Address Tuple or Address Lookup Table or Address Main table. If you want to clean the address then explore the BODS Address Cleansing service via MDM Enrichment architecture. If not then store it as it is.

3) DEBMAS and ADRMAS IDOC will be used, if you are using SAP PI in between SAP ERP & SAP MDM.

Thanks,

Shambhu.

[Former Member]

Hi Nitin,

To sum up you mean there has to be a main table with address details and linked to multiple customers.I feel this model will only apply to those 5000-6000 records for which you have to consolidate the data.

For the rest and the consolidated data ,the model should be more like the standard content for customer.

Eventually the records you consolidate and resultant clean data can be imported to that repository.

Either you can use 2 separate repository,in which one will only consolidate and provide data to the other.

The same can be used using 2 main tables.

For intial import tweaking MDIS.ini parameter as well as Inbound port parameters will be good for performance.

Thanks,

Ravi

[Former Member]

Hi,

Can you answer the following questions -

1) How you are going to use the Unique number generated in SAP MDM or how you are going to use the data stored in SAP MDM?

Since you do not have any key in the incoming data, you cannot send it back to the source system after match and merge.

If possible generate a key field in all source system. Use the concept of Remote System and Remote key in SAP MDM. With help of this you will be able to syndicate the unique number generated in SAP MDM back to the source system and your data will become standard in all source systems.

Read about Remote Systems Concept in MDM Console Reference guide - http://help.sap.com/saphelp_mdm550/helpdata/en/42/f3aaa5620e32c3e10000000a1553f6/frameset.htm

2) Have you explored the SAP Business Objects Data Services for doing analysis of your source data?

http://www.sap.com/solutions/sapbusinessobjects/large/eim/data-services/index.epx

Thanks,

Shambhu.

[Former Member]

Hi Ravi,

Thanks for the reply.

Nothing is primary key because we import the data into data manager,do the matching and merging and later we assign the number which will be unique.

There wont be more than 10000 records per file and also we recieve this file directle into the Ready folder of inbound port.

There are around 30 fields which are mainly address table details like Name 1,name2 ,search term1 and so on and one tuple which contains the customer number and source system.

Mine concern was like using 30 fields as combination to import would bring down the performane and stability of MDM

Please need your guidance

Thanks and Regards

Nitin

[Former Member]

Hi Nitin,

If the import is from SAP then the source should be ADRC table,this has multiple entries for some key field values which I guess you have made tuples.

I would suggest to you to goto we 60 transaction in SAP, and then set matching criteria based on the Idoc you are using.

Which idoc are you using here ADRMAS?

This can make it easier and stabler.

Thanks,

Ravi

[Former Member]

To add matching criteria with 30 fields should not be a worry,as working behind mdm combines all the fields to make a virtual unique field for the sake of comparision.

A test run of sample data will be good here.

It is equally important to note that you set appropriate Tuple update field and appropriate action.

Thanks,

Ravi

[former_member182007]

Nitin -

This is a good way, the only thing you break your data by considering your needs.

thx

Deep

[Former Member]

HI Nitin,

Yes one can use this as an option,it would not be a major concern as far as performance goes.

Also in this case breaking data in chunks of 10000-15000 will be a better idea.

But a better workaround would be to add some primary key,is'nt number primary key here?

Thanks,

Ravi